Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity Abderrahim Ouazzi, Stefan Turek Institute for Applied Mathematics, LS III, Dortmund University, Germany Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 1/19
18
Embed
Numerical Methods and Simulation Techniques for flow with ... fileNumerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity Abderrahim Ouazzi,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Numerical Methods and Simulation Techniquesfor flow with Shear and Pressure dependent
Viscosity
Abderrahim Ouazzi, Stefan Turek
Institute for Applied Mathematics, LS III,
Dortmund University,
Germany
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 1/19
Motivation of this work
The flow of granular materials
Example of applicationPharmaceutical Industry, Food Processing, Soil Mechanics ...
Granular material storage Couette flow
What about the viscosity !!?
From engineering point of view this material does not have anyviscosity!!
From mathematical and numerical point of view we are able to setthis type of fluid in the same range of flow with generalizedviscosity, since it exibits the same difficulties !?
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 2/19
Equations of motion
The general equation of motion for incompressible powders
Conservation of mass
DρDt
= ∂ρ∂t
+ ∇ · (ρu) = 0
D∗Dt
is the material derivative and u is the velocity vector
For an incompressible materialthe bulk density, ρ, is a constant thus
∇ · u = 0
The equation of motion
ρDuDt
= −∇ · T + ρg
with, T = S + pI, the deviatoric stress.
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 3/19
Constitutive equation
The constitutive equation is devoted to correlate between the deviatorictensor, S, and the velocity, throught the rate of deformationD = −1
2(∇u + ∇T u) , and assure the closure of equations.
Newtonian lawS = 2νoD
Power lawS = 2ν(DII)D, ν(z) = z
r2−1, r > 1
Schaeffer’s law (1987): For a powder a constitutive equation firstintroduced by Schaeffer (1997), which has to obey a
yield condition; ||S|| =√
2p sinφ, and
flow rule; S = λD
we use this correlation to obtain the constitutive equation
S =√
2p sinφ D
||D||
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 4/19
Generalized Navier-Stokes Equations
The generalized incompressible Navier-Stokes problem
ρDuDt
= −∇p + ∇ · (ν(p, DII)D) + ρg, ∇ · u = 0
If we define the nonlinear ‘’pseudo viscosity‘’ ν(·, ·) as a function ofDII(u) = 1
2D : D and p, then we can show that different materials couldbe ranged with different viscosity law including powder;
Power law defined forν(z, p) = νoz
r2−1
Bingham law defined for
ν(z, p) = νoz− 1
2
Schaeffer’s law (including the pressure) defined for
ν(z, p) =√
2 sin φ pz−1
2
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 5/19
New problems
In what follows we will show how to deal with the following problems
Discretization method:How to use nonconforming finite element methods for problemsinvolving rate of deformation tensor rather than the gradient !?
Nonlinear solver:How to apply Newton linearization technique for this highly nonlinearand irregular problem !?
Linear multigrid solver:In connection with the first two problems, how to keep the efficiency ofthe linear multigrid solver!?
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 6/19
Nonlinear Solver: Newton iteration
Let ul being the initial state, the (continuous) Newton method consists offinding u such that
∫
Ω2ν(DII(u
l), pl)D(u) : D(v)dx
+
∫
Ω2∂1ν(DII(u
l), pl)[D(ul) : D(u)][D(ul) : D(v)]dx
+
∫
Ω2∂2ν(DII(u
l), pl)[D(ul) : D(v)]pdx
=
∫
Ωfv −
∫
Ω2ν(DII(u
l), pl)D(ul) : D(v)dx, ∀v, (1)
where ∂iν(·, ·); i = 1, 2 is the partial derivative of ν related to the first andsecond variable, respectively.
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 7/19
New Linear Algebraic Problem
The algorithm consists of finding (u, p) as solution of the linear system
A(ul, pl)u + δdA∗(ul, pl)u + Bp + δpB
∗(ul, pl)p = Ru(ul, pl),
BT u = Rp(ul, pl),
(2)
where Ru(·, ·) and Rp(·, ·) denote the corresponding nonlinear residual termsfor the momentum and continuity equations, and the matrix A∗(ul, pl) andB∗(ul, pl) are defined as follows, respectively
〈A∗(ul, pl)u,v〉 =
∫
Ω2∂1ν(DII(u
l), pl)[D(ul) : D(u)][D(ul) : D(v)]dx. (3)
〈B∗(ul, pl)p,v〉 =
∫
Ω2∂2ν(DII(u
l), pl)[D(ul) : D(v)]pdx. (4)
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 8/19
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 9/19
Kernel function
The specific kernel function takes -1 or 1 in midpoints
The constants in Korn’s inequality for gradient, tensor and thestabilized tensor
NEL ‖uh‖G ‖uh‖T ‖uh‖ST
256 1.3 6.9 × 10−13
0.85
1024 1.4 1.4 × 10−13
0.90
4096 1.4 3.0 × 10−12
0.92
16384 1.4 6.1 × 10−12
0.93
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 10/19
Stabilized Rannacher-Turek Stokes Element
Remedy: Stabilized R-T FEM (Hansbo et. al)
The stabilization consists of adding the following bilinear form
∑
E∈EI∪ED
1
|E|
∫
E
[φi][φj ]ds (6)
for all basis function φi and φj with a weighted parameter s = s(ν), then thecorresponding matrix S is defined as:
< Su, v >=∑
E∈EI∪ED
1
|E|
∫
E
[u][v]ds (7)
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 11/19
Linear solverMultigrid for velocity and pressure simultanously:
Vanca-like block Gauss-Seidel scheme as smoother and solveradaptive step length control for correction step (with F-cycle)
macro-elementwise interpolation for grid-transfer
Reduced sparsity of the matrix S
Edge
Defect correction method:[
ul+1
pl+1
]
=
[
ul
pl
]
+ ωl∑
i
(
F + S∗|Ωi
B + δpB∗|Ωi
BT|Ωi
0
)−1[
Ru(ul, pl)
Rp(ul, pl)
]
For the preconditioning step only a part of the matrix, i.e. F + S∗, is taken
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 12/19
Newtonian caseIn this case the gradient and tensor formulation are equivalent; the efficiencyof the stabilized tensor discretization is checked by comparison with thegradient one on the flow around cylinder benchmark
gradient and stabilized tensor:similar behaviourtensor : the number of linear multigrid sweeps
increases with refinement; kernel funcion dominatesdecreases with the increases of the Reynold number; convectiondominates
This may explain why people from the CFD community did not pay attentionto this problem much more before!
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 15/19
Pressure dependent viscosityIn this case the nonlinear pseudo viscosity has the form ν(p, z) = Q(p)z
r2−1
The corresponding matrix of the linear problem
can no longer be fitted into classical saddle point problems;
Mδp(u, p) =
(
A B + δpB∗
BT 0
)
the solution is relative to the choice of imposing the uniqueness,since dim(null(Mδp
)) = 1
Efficiency of the solver:we increase the nonlinearity and list the numberof resulting nonlinear iterations and the averaged number of multigridsweeps per nonlinear iteration for both Newton and fixpoint
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 16/19
Schaeffer’s law
Schaeffer’s law:The time dependent equations are linearly ill-posedaccording to Schaeffer; in the Navier-Stokes equations, the pressureforce associated to the constraint div v = 0 can do no work. Bycontrast, the pressure force in equation of granular flow can do work,and for plane waves in certain directions, it does so.
Pseudo compressibility:This scheme is an effort to regularize theinstability in order to study the shear-band
average stress shear stress
t1 t2 t3 t1 t2 t3
The plot of the average stress and shear stress in a hopper shows thedevelopment of instability which leads to shear-banding.
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 17/19
OutlookWe have been able to develop the numerical methods and techniques tosimulate a dense granular flow, in futur we want to cover a wide range ofgranular materials
General equation of motion for a powder
ρDu
Dt= −∇p + ∇ ·
[
q(p,ρ)
||D− 1
n∇·uI||
(
D − 1n∇ · uI
)
]
+ ρg, with
Continuity equation∂ρ∂t
+ ∇ · (ρu) = 0, and
Normality condition∇ · u = ∂q(p,ρ)
∂p
∣
∣
∣
∣D − 1n∇ · uI
∣
∣
∣
∣
the yield condition q(p, ρ) is given by:
Powder properties Non-cohesive CohesiveIncompressible p sinφ p sinφ + c cos φ
Compressible p sinφ
[
2 − p
ρ1
β
]
p sinφρ1
β − C (p−ρ1
β )2
ρ1
β
Numerical Methods and Simulation Techniques for flow with Shear and Pressure dependent Viscosity – p. 18/19